With the increase of threats and high level of criminality, security remains a major public concern worldwide. Surveillance is one approach to address this issue. Through video surveillance, the behavior, activities, or other changing information, usually of people are monitored but which is often considered as a violation of the privacy.
The privacy in the surveillance video is safeguarded typically by encrypting region(s) of interest (ROIs) at the video capturing device (alternatively referred as camera) end either before or during video encoding. There are several limitations in encrypting ROIs at the camera end. The limitations are reduction in compression ratio (or increase in compressed bit rate) due to loss in correlation of ROIs, reduction in decoded video quality when performing encryption during encoding, static and fixed options of ROIs, necessity of multiple encoded streams if different ROIs is needed at different destinations, which significantly increases the encoded streams bit rate and higher complexity and power consumption at the camera end for providing the privacy control.
Hence, due to reduction in the compression ratio and multiple encoded streams, the bandwidth requirement from the camera for transmitting the encoded streams significantly increases. This calls for high bandwidth connectivity from the camera to a central server and terminals, which is not always available. And, the reduction in decoded video quality leads to poor visual experience and may create difficulties in detecting the regions of interest. Further, due to static and fixed number of ROIs at the camera end, the privacy levels supported are limited and fixed. The camera has to generate a plurality of video streams for each user based on authentication level of each user. Due to which there exists a lot of complication at the camera end and the resource consumption or computation level increases exponentially.
Another drawback in the existing technique is that, since the camera identifies the ROIs, masks the identified ROIs and transmits the masked ROIs to the user, if in case the user wishes to have a different authentication level, it is not possible.